Is my GPU dying?

[TommyTwommy]

My video card turns off when playing games or while being stress tested. What I mean under that is that my PC keeps running, the fans are spinning, RGB lights are on, but both of my screens go black (no signal)

Event log shows a warning "Display driver nvlddmkm stopped responding and has successfully recovered.", event id is 4101 (if that means anything)

 

Things I've tried so far:

• undervolting, underclocking and lowering power limit
• reseating the GPU multiple times,
• using different cables,
• reinstalling drivers (multiple times using DDU)
• memtest,
• using only one stick of RAM in different slots,
• turning off all unimportant background programs and services (like GHUB, Discord, NZXT Cam, etc.),
• different PCIe slot,
• older drivers

 

Is my GPU dying? It's an Asus RTX 3080 TUF, 3 years old. I never really used it that much. I cleaned it regularly and it is in a good ventilated case (temps never went above 65°C). It was also undervolted and running in "Quiet" mode since day one.

[emosun]

what power supply

[Razman]

I dont know about quality of your power supply but 1000w is more than enough, can you test the psu or the gpu in another system ? what you describing is usually one these two

also happens if you undervolt too much

[TommyTwommy]

15 minutes ago, emosun said:

what power supply

Revolt Pro 1000W. It's a very high quality power supply and more than enough power for an RTX 3080

 

  

7 minutes ago, Razman said:

I dont know about quality of your power supply but 1000w is more than enough, can you test the psu or the gpu in another system ? what you describing is usually one these two

also happens if you undervolt too much

Yes, this is the only thing left to try but I don't have another system at my place. I'd have to give it to a friend, but none of them have a PSU that's powerful enough to test it.

 

I am currently trying to undervolt but for some reason my voltage curve is locked and I can't move it..

[--SID--]

4 minutes ago, TommyTwommy said:

Revolt Pro 1000W. It's a very high quality power supply and more than enough power for an RTX 3080

It's decent but very high quality is a bridge to far for a Seasonic Focus based PSU.

 

With how many seperate cables is the PSU connected to the GPU?

 

Are there extensions on the PSU cables?

[TommyTwommy]

3 minutes ago, --SID-- said:

It's decent but very high quality is a bridge to far for a Seasonic Focus based PSU.

 

With how many seperate kabels is the PSU connected to the GPU?

 

Are there extensions on the PSU cables?

It's 2 separate cables. Yes, I did have extension cables, but I removed them since.

I can normally use my PC, but after playing a game for 1-30 minutes it will turn off.

 

I mean it's an A tier power supply. It shouldn't be causing issues. I'm sure there's other better power supplies, but this one is more than decent

[leclod]

Did you check your gpu temps (and by the way cpu temps, all temps) with HWinfo64 (sensors) ? (your gpu might have more than just one temp sensor)

I'm not sure, did you try the gpu (and by the way the whole system) at stock settings ?

 

1 hour ago, TommyTwommy said:

I am currently trying to undervolt but for some reason my voltage curve is locked and I can't move it..

How do you do that, with Afterburner ?

[freeagent]

Those were the symptoms that I noticed when my XFX GTX295 turned into a GTX265

 

I would RMA the card if you can.

[TommyTwommy]

12 hours ago, freeagent said:

Those were the symptoms that I noticed when my XFX GTX295 turned into a GTX265

 

I would RMA the card if you can.

The 36 month warranty expired a little over a month ago

 

21 hours ago, leclod said:

Did you check your gpu temps (and by the way cpu temps, all temps) with HWinfo64 (sensors) ? (your gpu might have more than just one temp sensor)

I'm not sure, did you try the gpu (and by the way the whole system) at stock settings ?

 

How do you do that, with Afterburner ?

All temps are fine. For all components and cores. Yes I was using MSI Afterburner.

[leclod]

You didn't answer : did you try the gpu (and by the way the whole system) at stock settings ? maybe your undervolting is unstable somehow.

 

Also you could try a full reinstall.

You could disassemble the gpu and check/replace the thermal pads, if you're handy.

Edited December 17, 2023 by leclod

[TommyTwommy]

1 minute ago, leclod said:

You didn't answer : did you try the gpu (and by the way the whole system) at stock settings ? maybe your undervolting is unstable somehow.

 

Also you could try a full reinstall.

You could disassemble the gpu and check/replace the thermal pads if you're handy.

Yes, I returned the voltage settings to stock and tried running at both the stock clocks and underclocked.

 

I didn't do a full reinstall yet, it's going to be the last thing I do. I hate doing reinstalls because I always forget to back something up or forget how I set something up in the first place so it takes me months to get it back to the state I want. I have a lot of small modifications and tools installed for Windows 10 (half of which I forgot the name for). Many settings from tons of apps, databases, etc. It all resets when you reinstall windows, so I really want to avoid doing it. Thing is I didn't change or install anything new when this started happening so I doubt a reinstall will fix this anyways.

 

I'd rather try the card in a different system to confirm it's the card.

[leclod]

31 minutes ago, TommyTwommy said:

I didn't do a full reinstall yet, it's going to be the last thing I do. I hate doing reinstalls

Don't you have a spare drive to install an OS to with no strings attached ?

Edited December 17, 2023 by leclod

[TommyTwommy]

32 minutes ago, leclod said:

Don't you have a spare drive to install an OS to with no strings attached ?

I might have spare drive.. I could try that yeah, good idea

[TommyTwommy]

11 hours ago, leclod said:

Don't you have a spare drive to install an OS to with no strings attached ?

I ran furmark at highest settings for 30 minutes on a new installation while overclocked and at highest power limit and it didn't crash, so I was like oh ok it was the windows installation. But I'm now also 20 minutes into the same stress test on my actual windows installation and it's not crashing either. I'm so confused.

Usually it would crash within first 2 minutes of furmark everytime I'd run it.

[LOST TALE]

I wouldn't undervolt. if you kept temps under 65 and it crashes at normal voltage, it's a defective gpu.

I've had one like this. It had seasons and would magically become reliable again. and times where i'd run it at -50% power praying it works so I dont have to buy at 10x prices during the mining rush.